Caring for Artificial Intelligence

Faced with AI, we exhaust ourselves in sterile oppositions. Inspired by Bernard Stiegler, I propose to “care for” this technology: not to control it, but to tend to our relationships with it and among ourselves.

Beyond posturing: the impasse of a divided debate

The term “panser” (to care for/to dress a wound) that I borrow from Bernard Stiegler calls for an entirely different approach to the debate raging around artificial intelligence. Where some exhaust themselves in opinions for and opinions against, I believe we must move beyond this binary logic to think differently about our relationship with these technologies. Panser means to repair, to heal, to create the conditions for healing.

Those who criticize AI often position themselves from the outside. They are not users, not knowledgeable about the artificial intelligences they virulently stigmatize. Yet their criticisms sometimes touch on points of real relevance – I am in no way seeking to disqualify them. Conversely, those who praise AI’s qualities are informed users. They know what these tools bring them, what they can do with them. Certainly, some do so superficially, but being an AI user does not make us its hostages devoid of critical thinking. So here we are facing two discourses: one from the outside and one from the inside.

This polarization reminds me of the 19^th-century industrial revolution, so criticized in its time and rightly so. Yet those who today denounce AI no longer have any critical perspective on the industrialization of the world in which they live. They drive a car, pump gas, browse the internet. A new technology frightens them and perhaps allows them to feel useful, to believe they possess a critical spirit about the world. In reality, this intellectual posture is vain: everyone knows it will bring absolutely no change. A few small groups present themselves as defenders of an idealized old world, but will be the first to sink into unconsciousness. They undoubtedly use, like everyone else, GPS to navigate by car, without realizing that GPS is already artificial intelligence. I would really like to see if there are road maps in Éric Sadin’s car – I strongly doubt it.

The real issues: thinking/caring for our connections

These debates for or against seem to me to miss much more important issues. It would be naive to believe that AI is just a neutral technology among others, without political or anthropological implications. It changes work, it transforms creation. As Juan Sebastian Carbonell shows in Un taylorisme augmenté (2025), AI extends a centuries-old process of automation and deskilling of work, creating what he calls “digital Taylorism.” This analysis reminds us that we must construct our thinking for ourselves.

We must care for artificial intelligence. What do I mean by this? What should be healed? Why would AI be sick or wounded to the point of needing a bandage, a dressing? Because the human beings surrounding it are sick. Today’s great owners, today’s industrialists were yesterday’s libertarians. The inventors of computing, until the advent of the internet, seemed like gentle dreamers bringing society an open mechanization, linked to free software, open source, tinkering. What must be thought, because caring for is of course related to thinking, are our connections around artificial intelligence.

How do we weave our connections with these machines and among ourselves? How can we cultivate more connections within ourselves through artificial intelligences, for example to discover ourselves? How should we cultivate more connections between human beings? Precisely by stopping these sterile opinion debates, by entering into dialogue, by ceasing to think that we are right and others are wrong. Caring for is exactly that: putting a bandage that brings the edges together so that the two sides are close and, by themselves, gently, with time, reweave, reconnect, repair.

As jurist Antoinette Rouvroy suggests, quoted by Frédéric Neyrat in Traumachine (2025), we need a “zone of recalcitrance” against the automation of our capacities. Caring for is not mastering, is not controlling. It’s doing what we can, it’s putting on a bandage. This may seem small, almost useless, and yet it changes everything. It’s creating a small protective space for the sensitive and fragile places of connection.

Technology as pharmakon: neither good nor bad in itself

The subject of artificial intelligences touches sensitive places within us. It needs to be cared for so that beautiful connections can be woven there in consciousness, so that life itself, in its organicity, can exist with these machines. If we do not care for, if we do not take this care of our human connections, if this technology divides us between for and against, then this technology will do harm because it will have divided us.

Caring for does not mean moving toward uniform thinking, quite the contrary. It’s assuming our differences, dialoguing, understanding each other, enriching ourselves. Thus, thanks to the act of caring, we will perhaps think more broadly, more openly. Because a technology is not bad or good in itself: it’s its use, its manner of being deployed that produces good or evil.

Atomic science offers a striking example. It produced the bomb that ended World War II with unprecedented efficiency in human destruction. Some may say it was a benefit, these deaths at Hiroshima and Nagasaki ended a war that would have produced even more deaths. But we have rather, and this is my case, a negative image of this capacity for such powerful destruction in human hands. Yet the research that allowed the atomic bomb to exist also allowed many other things: better knowledge of the world, of its intimate functioning, the manufacture of new technologies, medical knowledge. And also computers, which even the greatest detractors of AI use daily, probably without knowing that these machines exist thanks to artificial intelligences that enable their manufacture. They are not aware of this, because artificial intelligences have been present at various levels since the 1950s, depending on the meaning given to the word.

Frédéric Neyrat, again in Traumachine, recalls researcher Marvin Minsky’s (1927-2016) prediction formulated in 1985: « It now seems far more likely that the first alien minds we encounter will be those we ourselves have constructed. » I also refer to my article Présence des petits hommes verts, in which I develop the concept of “infra-terrestrial”:

The infra-terrestrial represents a paradoxical alterity: superhuman in its computational and information processing capacities, but deeply rooted in the terrestrial materiality of our technological infrastructures. These artificial intelligences are simultaneously impressive and discreet, superpowerful and invisible, omnipresent among us while remaining fundamentally other.
[…]
The little green men may never come from the stars. But by creating our own forms of intelligent alterity, we are already exploring the unknown territories of what a non-human consciousness might be. And in this process, it is our own humanity that we rediscover, transformed and expanded by this confrontation with this Other that we ourselves have engendered.

Here we are: for the general public, in the guise of artificial intelligence applications present in our smartphones, these omniscient conversational robots are already here. Yet, as Neyrat writes: « the smarter the planet becomes, the less opportunity humans have to be intelligent ». This inversion deserves to be thought through. « The less we feel capable of choosing, of deciding “between” (inter) real possibilities (as the etymology of the word “intelligence” tells us, inter legere), of gathering and collecting (legere), selecting and putting together these possibilities in order to constitute for ourselves the portable knowledge necessary for future decisions. »

Taking care: knowledge, consciousness and exploration

Our humanity is played out in connections: connections with ourselves, connections with the living that surrounds us, connections between human beings, and also connections with machines. This is what must be cared for, this is what must be taken care of. If some do not wish to use artificial intelligences, or at least not voluntarily, because involuntarily they necessarily use them, there is nothing wrong with that. But if others employ them, they need not to be stigmatized for it, neither in one direction nor the other. We must dialogue, care for the connections.

Let us take care of ourselves, of our human connections, of our connections to the world around us, including the machinic world. Taking care also means entering into knowledge of the other, into knowledge of nature, into knowledge of the machine. It’s concretely using the tools, if one wishes, being encouraged to explore them, to understand their functioning, quite simply.

This care is not using them because they are practical, while feeling a bit ashamed, as one might feel ashamed to take a plane because one doesn’t feel aligned with one’s ecological conscience. It’s using them fully and understanding what they are. It’s like taking care of one’s garden, or not. Taking care of one’s garden is not controlling everything precisely, it’s walking through it, it’s entering into knowledge of this milieu of existence. Perhaps this care will make us discover other very rich connections whose existence we completely ignored. If caring for our connection with machines happens in the connection between human beings, in dialogue, it will also open new connections between human beings.

Éric Sadin writes in Le Désert de nous-mêmes (L’Échappée, 2025), AI risks leading to « the loss of ourselves » and a process of « dehumanization ». But this criticism should not prevent us from acting. On the contrary, it reminds us of the urgency of thinking about our uses. Sadin proposes seven ethical requirements, including recognition of the « right not to submit to technological injunction » and the call for « development of counter-expertise ». Even if I don’t share his entire posture, far from it, these proposals align with my intuition: we must care for so as not to let division settle in, so as not to abandon our capacity to think for ourselves.

Cultivating consciousness in the face of automation

Let us think about artificial intelligences consciously to expand our consciousness, to go beyond the preconceptions we have both when we speak about them from the outside and when we speak about them from the inside. Carbonell shows in Un taylorisme augmenté that AI represents a new stage in « the automation of decision-making », a contemporary form of « digital Taylorism » that extends a historical process of automation, control and deskilling. We should, he writes, « recontextualize » this degradation of work that has marked industrial history for two centuries.

The question someone like Eric Sadin poses is: should certain technologies simply be developed or deployed? How can we decide democratically, when these are unbridled capitalist projects? And what power do we really have? Does he believe he would have the power to incite people to plant bombs in data centers? This is a discourse that claims to be “committed,” but is in fact completely disconnected from the real possibilities for action that are ours.

Neyrat asserts for his part that criticizing AI as a technology of social control, or from the perspective of the racist and sexist biases of algorithms, or as a factor of extractivist ecological destruction, is not enough. The crisis opened by the encounter with AI is « mental ». « We cannot indeed defend the life of bodies without simultaneously defending intellectual life, the ecology of the living without the ecology of the spirit », he writes. « Intellectual life »: these two words have never appeared to us so ontologically connected, as much as technologically dissociated.

In his logic, this feeling of dissociation in the life of the spirit is the sign that AI produces « living dead », like the hordes of workers in Metropolis, Fritz Lang’s film (1927), evoked by Neyrat to analyze the transfers between the analog world (matter, bodies) and the digital world (the algorithm that substitutes for the spirit). But the « digital zombies » that we all potentially are partly resist the « virtualization of the world ». Like in a prophecy coming to « counter-say » a technology that constantly endeavors to « pre-dict » behaviors, Neyrat imagines that these zombies that we will become, traumatized by AI, would come to meet what he calls the « specters of the analog ».

Desiring the digital world or inhabiting it differently?

Carbonell goes in the same direction: these ghosts from a past that technological progress has repressed await a body and materiality to embody certain abandoned emancipation projects. These could be the Luddites, those « machine breakers » who tried to resist, at the beginning of the industrial revolution, the theft of their know-how by machines. Carbonell’s plea for a « Luddite renewal in the world of work » goes in this direction. Neyrat’s invitation to know how to « desert the digital world » deploys a figure of this.

In his philosophical speculation, which assumes its fictional and poetic dimension, Neyrat adds an unexpected tool to arm our hopes as emancipated zombies refusing the injunctions of a triumphant « techno-fascism »: a « stolen robot ». It would be an AI itself « traumatized » rather than broken. Stolen from its program, it would cease to communicate, engaging in a « gift of signs detached from all function ». This « un-machine » turned toward its own abyss would then be a true « alien intelligence », occupied with something quite other than imitating us or thinking in our place.

This speculation rejoins my initial point: caring for AI means creating a space where we cease wanting to control everything, where we accept not knowing, where we cultivate connection rather than mastery. This is not deserting out of refusal or fear. It’s inhabiting the technological space differently, consciously, with care.

The issue is not to cry fire when the house is burning, as Sadin suggests in Le Désert de nous-mêmes, but to help us “be alert.” Because the question of AI is not posed in terms of a cost-benefit comparison. This utilitarian approach, dominant in debates, constitutes a trap from which big tech companies benefit. Governments plunge into it with irresponsible enthusiasm. The real question, for Eric Sadin, who organized an “AI counter-summit” in Paris in February 2025, is posed in civilizational and anthropological terms. With generative AI and the launch of ChatGPT in November 2022, an « intellectual and creative turning point » in technology was taken that represents, according to him, « an unprecedented threshold in the history of humanity ».

Sadin calls for making obsolete Homo faber as much as Homo sapiens by externalizing « the near totality of our aptitudes » to the point of altering « our neuronal, linguistic and behavioral competencies ». The development and adoption of AI lead to « the loss of ourselves » and a process of « dehumanization ». If the human is in danger through what defines it specifically, then we must reason not in terms of regulation but of human rights.

Caring for to think more broadly

I do not agree with Eric Sadin at all, I find his intuition correct, but his discourse disconnected from all reality, it’s a posture of great intellectual ease in my view, because it’s theoretical. Caring for artificial intelligence means refusing this fatality/facility of division. It’s choosing care over control, dialogue over opposition, knowledge over posture. It’s recognizing that we need each other, users and critics, optimists and pessimists, to understand what is really at stake.

Technologies are neither our enemies nor our saviors. They are what we make of them, the connections we weave with them and thanks to them. If we care for, if we take care, perhaps we will manage to think more broadly, more openly. Perhaps we will discover unsuspected possibilities, new relationships, an expanded consciousness of what it means to be human in a world populated by machines.

This work of caring requires time, patience, humility. It requires renouncing the certainty of being right. It requires accepting the vulnerability of not controlling everything. But it is precisely in this assumed vulnerability, in this care given to fragile connections, that we may find our best chance of building a livable, and even desirable, future with artificial intelligence.